Introduction

The global prevalence of overweight and obesity is high [1, 2] and South Africa is no exception [3, 4]. Worldwide no country has been able to turn around the rising numbers [2, 5], and the increase has been particularly steep in South Africa [3, 4]. The South Africa Demographic and Health Survey (2016) reports an overweight and obesity prevalence of 31% among South African men and an even higher prevalence of 68% among South African women [4]. The prevalence of noncommunicable diseases (NCDs), such as hypertension and diabetes, is increasing annually in low- and middle-to-high-income countries, including South Africa [4, 6]. Comprehensive and effective corrective measures are needed to address these trends.

Due to urbanization, access to a wide variety of ultra-processed foods and non-alcoholic beverages (henceforth referred to as drinks in this paper) has increased remarkably in the South African market since the early 1990s [7, 8]. Several reports have demonstrated a link between high consumption of ultra-processed foods and drinks and the development of obesity and NCDs [9–11]. Urbanization has resulted in a shift from traditional diets to consumption of ultra-processed foods [8, 12, 13] made from multi-ingredient formulated mixtures containing little if any whole foods, such as ready-to-eat refrigerated processed meat (known as polonies), carbonated drinks, biscuits, and breakfast cereals, and that are typically high in sodium, sugar, saturated fats, and calories [14]. In a study conducted in Soweto, South Africa, most adolescents reported a high frequency of fast food consumption per week, with sweets, crisps, and soft drinks accounting for more than 65% of the total items consumed [15]. In other studies South African adults have reported very high sodium intakes due to consumption of processed food, including ultra-processed food [4, 12].

A simple, easily understandable food labeling system could assist consumers in identifying unhealthy food options amid the wide variety of packaged products available in markets and steer consumers away from them. Simplified FOPL has a role to play in addressing these health concerns by empowering consumers with information about products’ nutrient content. Front-of-package labeling (FOPL) is a practical tool that empowers consumers to make informed food choices at the points of purchase and consumption [16, 17]. Several international studies have demonstrated the usefulness of FOPL in assisting consumers to identify unhealthy food products and discouraging selection of products identified as unhealthy [18–21].

FOPL systems range from reductive labels, which mainly summarize the nutrition information in the back-of-pack Nutrition Information Panel and present it on the front of the package, for example, Guideline Daily Amounts (GDA), to interpretive labels, which evaluate the nutritional quality of food products and present the information with icons or symbols, for example, warning labels and health logos [22, 23]. Evidence indicates that interpretive food labels are more effective in directing food choices than their counterparts [16, 20, 24].

Warning labels are interpretive FOPL systems that are implemented in countries such as Chile and Israel to highlight products that are excessive in energy, saturated fats, sugar and sodium [25, 26]. This labelling system aims to discourage purchasing and overconsumption of unhealthy products by flagging products which contain excessive nutrients of concern in a simple, visible and easily understood manner [20, 27]. Highlighting nutrients associated with NCDs may increase risk perception, foster easy identification of unhealthy products, and discourage their purchasing and overconsumption [24, 28, 29]. Consumers have limited shopping time [30, 31] and warning labels that are conspicuous serve as a means to quickly identify unhealthy products within a short period [32].

Recent studies in France, Australia, Brazil and Chile indicate that, in comparison with other systems, warning labels are more effective in decreasing the intention to purchase unhealthy foods and drinks [18, 28, 33] and in helping consumers identify less healthy foods and drinks [34, 35]. Evidence from the tobacco industry also points to the potential of health warning labels to reduce use of a harmful product [36, 37]. The use of health warning labels with images alongside text on tobacco packages has been associated with increased smoking cessation [39] and attempted cessation [38]. Warning labels increase risk perception through eliciting negative emotions such as fear, discomfort and worry and thus are effective at reducing purchasing intention [32, 38, 39].

Although studies revealed the successfulness of warning labels in dissuading consumers from purchasing unhealthy products and decreasing the perceived healthfulness of unhealthy products [20, 40], no significant differences in the mean nutritional composition of purchased products were noted in certain cases [41]. Warning labels by their binary nature (warning vs no warning) and because they do not state nutrient amounts are reported to be less informative, especially among the highly educated groups [21, 42]. The nutrition facts panel at the back of the package could however be used to complement the front of pack nutrition information.

Nutrition warning labels highlight excessive nutrients of concern and often include the text “high in” or “excess,” warning consumers that the levels of those nutrients are above health recommendations [25, 27]. The text is usually enclosed in familiar shapes such as a triangle or octagon associated with some form of risk and may be accompanied by text depicting danger or caution [25, 43]. Repeated exposure to familiar shapes such as the stop sign or triangle attracts consumer attention [44] and increases danger or risk awareness [43]. Deliza et al. [45] found that participants located black triangles and black octagons most quickly on product packages in comparison to other shapes and FOPL formats. According to the Communication–Human Information Processing (C–HIP) Model, warning labels that are attended to, are well understood and increase risk perception, influence behaviour and may ultimately lead to habit change [46–48]. Use of color such as black or red, which is associated with ‘stop’ or danger [49]; bold font, which is attention grabbing [43]; and bigger label size enhance the labels’ visibility and captures attention [43]. In a study by Cabrera et al. [50] a red stop sign with the text ‘excess’ was associated with the least perceived healthfulness and black was reported more visible. These attributes enhance the warning labels’ effectiveness in informing consumers and steering them away from unhealthy food [51]. In addition pictorial images and icons further increase visibility [52] and simplify understanding of warning labels across various sociodemographic groups, including low-socioeconomic, low literate groups and children [18, 32]. Images are easy to understand and are more easily retained in memory [53].

The current nutrition panel currently used in South Africa is on the back of the packaged, is complicated and not well understood due to the terminology and the difficulty in interpreting numbers used [30, 54]. A study by Jacobs et al. [55] revealed that South African consumers previously expressed a need for an FOPL system that communicates nutrition information in a simple manner. South Africa as a country has not yet implemented any FOPL system The success of warning labels [18, 34, 35] in other countries created an opportunity to develop and test a a warning label that could effectively warn and modify South African consumers’ purchases of unhealthy products. This study aimed to fill this gap by probing South African consumers’ responses to warning labels and exploring their views on features that could enhance or diminish its effectiveness. The study explores (1) consumers’ perceptions of warning labels and (2) consumers’ views on design features that could influence the effectiveness of a warning label.

Materials and methods

Sample

The market research company recruited participants through their existing database using a recruitment questionnaire developed by the research team (MB, LST, NM, TC, LM and RS) (S1 File). Data were collected from 12 focus groups of 8 to 10 members each, with a total of 113 participants taking part in the study. A sample of forty (40) participants was targeted per province to represent the various sociodemographic strata. The sample included adults primarily responsible for household food purchases purposefully selected using quotas stratified according to gender (male or female), age (18–29 years or 30–50 years), literacy (no literacy, low literacy, or literate), income (low or middle-high), urbanicity (urban or rural), and geographic location (Gauteng, KwaZulu-Natal, or Western Cape provinces) (Table 1). The focus groups were homogenous with each group consisting of participants with similar sociodemographic characteristics. We classified household income levels as low if less than R1,600 (approximately USD 100) per month and middle-high if R1,601 and above. We defined no literacy as adults with no formal schooling, low literacy as adults who had passed grades 1–6, and literate as adults who had passed grade 7 or higher.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Socio-demographic characteristics of participants (n = 113).

https://doi.org/10.1371/journal.pone.0257626.t001

Label design

A design agency created several warning label prototypes for testing on South African consumers following a detailed design brief based on the latest literature [36, 43]. The designer was briefed to create an effective warning label using shapes, text and icons that would increase South African consumer’s identification of the presence of unhealthy nutrients in foods and encourage healthy choices. The intent was also to ensure that the labels would be effective across diverse sociodemographic groups in South Africa. A committee comprised of experts in nutrition, health, health promotion, economics, communication, and media systematically rated each South African design to narrow down the range of options. Based on recommendations by the expert committee, we first tested black triangles on a white holding strap with the word “warning” (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Image of the warning label designed for South African consumers.

https://doi.org/10.1371/journal.pone.0257626.g001

Each triangle contained the text “high in” and icons depicting the nutrient of concern present, sugar, saturated fat, and/or sodium (Fig 2). The warning labels were superimposed on the front of four different food and drinks packages i.e. chips/crisps (square packet), fruit juice (bottle), yoghurt (yoghurt container) and cereal (rectangular box). A product high in all three nutrients would show three corresponding triangles with the relevant icons. We used this original warning label to evaluate consumers’ perceptions of warning labels.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

The original warning label with black and white triangles tested in the first part of the study (images from left to right: A) crisps, B)100% fruit juice, C) yoghurt and D) cereal).

https://doi.org/10.1371/journal.pone.0257626.g002

The second part of the study compared different design elements of the warning label (Table 2 and S1 Fig) to determine the preferred version.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Design elements tested.

https://doi.org/10.1371/journal.pone.0257626.t002

We tested 32 design element options of the warning label. This included different types of icons for each nutrient, symbol shapes (triangles versus octagons), symbol colors (black versus red), holding strap colors (black versus white), warning devices (warning text only, warning text with an exclamation mark, black no warning device and red no warning device), text fonts (uppercase versus lowercase letters), and label sizes (Table 2). Other countries have similarly tested a number of prototypes [59].

Results

We extracted 5 themes and 16 subthemes from the data (Table 3 and S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Themes and subthemes.

https://doi.org/10.1371/journal.pone.0257626.t003

Elements perceived effective as warning

Fig 2 shows the design features participants perceived as 1) attention grabbing, (2) effective as a warning against unhealthy foods and drinks, and 3) likely to influence their purchasing behaviours. They considered that a black triangle on a white background (holding strap), placement in the top right corner of the front of the package, and uppercase letters improved the visibility of the warning label. Participants viewed a combination of the word “warning” and a triangle with an exclamation mark as signal for danger. They also preferred a larger label compared to smaller sizes (Fig 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Design features that appealed to participants as depicting warning.

https://doi.org/10.1371/journal.pone.0257626.g003

Discussion

This study found that adult South African consumers had a positive attitude toward warning labels on ultra-processed foods and drinks. Study participants felt that warning labels are easy to understand, provide important nutrition information, and could save shopping time. Our participants generally found warning labels visible and expressed that the warning labels increased risk perception. These elements are associated with the efficacy of warning labels [43]. An easily understandable and readily visible warning label attracts attention [43] and increases the likelihood of label reading, use and behavior modification [31, 43, 46, 63].Previous studies in Uruguay and Chile reported acceptance of warning labels among their participants and found the labels easy to interpret [59, 64]. Findings from quantitative studies support that consumers understand warning labels better than other label formats [27, 35, 65–67]. The advantage of easily understood warning labels is a wider coverage of consumers, even those without access to individual dietary counseling by health professionals.

This study also found a common perception that the summarized nutrition information on warning labels would easily help consumers identify unhealthy products at the point of purchase. Warning labels display only the nutrients in excess [27, 66], therefore consumers can quickly [68] and easily identify unhealthy products [18, 32]. This was supported by other participants in this study who were of the opinion that the warning labels would save them time. Consumers often have time constraints and warning labels that cut through the noise on the product packages would be beneficial [21, 69].

The use of familiar shapes [20, 52] and colors [70, 71] associated with danger increases the effectiveness of warning labels. As mandated by regulatory authorities, shapes and colors are used successfully in other industries, including the tobacco industry and segments of the food industry, to raise awareness of dangers of consuming those products [36]. However shape association with danger is culturally learned [41] and not universally interpreted in the same way [50]. In this study South African consumers perceived the black triangle in the warning label as depicting danger. A triangle is commonly used in construction sites, workplaces and road signs in South Africa to signal danger, so it is not surprising that consumers in our study in all sociodemographic groups favored it over the octagon shape. Consumers in Brazil [19] similarly related triangles with danger, but in other countries, such as Chile [59, 72] and Israel [26], consumers perceived the octagon as communicating a warning best. In the latter countries, the octagon stop sign was better understood than other shapes. These results indicate the need for each country to investigate its own context specific shape preference to improve the efficacy of the warning labels.

Although some consumers in this study, particularly those of low-socioeconomic status, perceived red as signaling danger and attention grabbing than black, black was deemed better as it contrasted with the colorful product backgrounds. Similarly, Cabrera et al. reported that black signs were easier to locate on colorful packages than red ones [50]. Color, particularly red, increases the visibility of FOPLs, increases risk perception, and influences behavior change [73]. However, visibility against the competing background is also important for warning label effectiveness [43] and a black label against a white background stood out more to our participants. In Israel, however, consumers preferred the red label [74]. Placement of the black triangle on a white background also could have improved the triangle’s salience. Policy maker could further explore inclusion of a white background as a means to make the label stand out by s. The findings of this study also confirm that consumers preferred the black triangle on the white background (i.e. holding strap) than a black triangle on a black background or without any holding strap. Chile [72] and Uruguay [75] use black and white warning labels while Israel [26] uses red and white circles.

Text and icons are among the elements that our participants thought communicated nutrition information clearly and simply. Words such as “warning,” “caution,” “excess,” and “high in” increase risk perception and improve effectiveness [38, 43, 50]. Similarly, in this study participants pointed out that the word “warning” on labels made it clear that they were being warned about the danger associated with consumption of the product. Evidence shows that icons that summarize nutrition information are beneficial to less literate groups [32, 76, 77], which is particularly important for South Africans with lower literacy levels.

An advantage of warning labels is their potential to influence consumers’ purchasing behaviors [24, 28]. Quantitative studies evaluating implementation of warning labels show that chocolate and cookie sales decreased by 8.0% and 1.2%, respectively, in Chile [78] and decreased expenditures on sweets and desserts in an online simulated environment in Uruguay [41]. In this study consumers indicated their intentions to reduce purchases of unhealthy foods with warning labels, particularly the products they like. Some even expressed an intention to stop consumption of those products altogether over time. Similarly, in a qualitative study in Brazil consumers stated they would continue consuming products with warning labels but at reduced frequency [79]. The intention to reduce consumption of ultra-processed foods and drinks is in line with the South African Food Based Dietary Guidelines, which recommend consuming fats, salt, and sugar sparingly [80]. Other participants in the current study however perceived that they would not be deterred by the warning label in line with the Health Belief Model which posits that low risk perception does not elicit behaviour change. Familiarity with products also decreases the effectiveness of the warning labels [55, 81].

An experimental study in Uruguay reported that warning labels impacted children’s food choices much better than the MTL [18]. In Brazil, de Morais Sato et al. found that parents perceived that easily read and understood warning labels would help their children independently identify unhealthy food products and would increase their autonomy in healthy food choices [79]. Our participants agreed that children would benefit from simple warning labels that encourage them to make healthy choices. This observation is critical, because reducing childhood obesity by reducing their consumption of ultra-processed foods and drink is urgent. Ultra-processed foods are often marketed as convenient and palatable and front-of-package warning labels steer attention towards the unhealthiness of the products.

Our participants recognized several design features that could potentially enhance the effectiveness of warning labels, including a black triangle on a white background (holding strap), location of the warning label in the top right corner of the package, and text in uppercase letters for clear visibility. They noted that a combination of the word “warning” with a triangle containing an exclamation mark on warning labels could further effectively alert consumers to potential health risks. They also preferred a larger warning label rather than a smaller one (S1 Fig).

A strength of the study is that it considered views of consumers from diverse sociodemographic backgrounds and offered them a combination of images of foods and drinks that are perceived as healthy and unhealthy to minimize preconceived notions about the nutritional quality of the products. As with any qualitative study, the sample is not representative of the entire population and the findings cannot be interpreted statistically. Understanding labels is important for influencing behaviors, and future research should investigate the influence of the warning label on purchasing behavior in a real shopping environment.

In conclusion, our results from focus groups in South Africa suggest that a policy mandating nutrition warning labels on unhealthy packaged foods could improve consumers’ understanding of health risks and help them identify unhealthy foods and drinks. Certain design elements, such as color (black), shape (triangle), text (warning and ‘High in’), use of exclamation mark and contrasting white background could enhance a label’s ability to increase perception of the risks of consumption of unhealthy products.

Supporting information

Acknowledgments

Special thanks to Emily Busey, graphic designer for superimposing warning labels onto food packages; to Alexey Kotov for his review of the discussion guide and for his support to the focus group discussions. We thank School of Public Health at the University of the Western Cape and the DSI/NRF CoE in Food Security (UID 91490) for administrative support.